Summary The Bell 204B helicopter (registrationC-GEAP, serial number2063) operated by Gemini Helicopters Inc. was involved in forest fire suppression at Bonaparte Lake, British Columbia. At about 1105 Pacific daylight time, the helicopter departed a staging site, eastbound, slinging an empty water bucket on a 100-foot longline for the first mission of the day. Shortly after take-off, the helicopter emitted a high-pitched, oscillating sound. The flight path and behaviour of the helicopter were normal as it went out of view over some trees. Immediately thereafter, there was the pronounced sound of main-rotor blade slap, followed by the sounds of impact with the trees. The helicopter struck the ground just short of reaching a small clearing adjacent to a fire road about nautical mile southeast of the staging site. A post-impact fire ensued, which destroyed the helicopter cabin area and melted the transmission casing and pylon structure. The main driveshaft assembly remained attached to the engine and transmission input quill assembly. The water bucket was found in a tree, detached from the longline, on the approach to the accident site. The longline was wrapped around another tree and lay in a direct line to the helicopter; other than the metal parts, the top 15feet of the artificial fabric line was burned away. British Columbia Forest Service crews attended the site and extinguished the fire. The pilot was fatally injured. Ce rapport est galement disponible en franais. 1.0 Factual Information 1.1 Operations Background On Wednesday, 13 August 2003,1 the occurrence pilot completed the daily flights at the Bonaparte Lake site and returned to Kamloops, British Columbia, where the engine and transmission were to be changed; both were time expired and due for an overhaul. The total airframe time was 11538.8hours. On 13and 14August, the engine and transmission assemblies were changed by company maintenance personnel. Honeywell (Lycoming) engine assembly modelT5311B, part number1-000-080-11, serial numberLE-08228, was installed. This engine had a total time since new of 7398.1hours and had been overhauled to a zero-time condition. Bell Helicopter Textron transmission assembly part number204-040-009-087, serial numberB12-601, was also installed. This transmission had a total time since new of 6808.2hours and 396.5hours since overhaul. The new components were run for the first time on the evening of 14August, and the following morning, an engine on-ground power check (partial power) was performed. The power plant was operating at 1.4percent below the calculated N12rpm required, and N1take-off trim adjustment of 3/8of a turn was required to increase N1to the calculated figure. (Consultations with industry personnel indicate that N1take-off trim adjustments are not an unusual requirement.) An entry was made in the aircraft journey log indicating that all work done was certified, pending a satisfactory test flight, including an N1topping check and a power check. A test flight was performed and engine operating parameters relevant to the power check were recorded in the aircraft journey log on 15August2003. The data was plotted against the engine operation and functional-check charts, and results were also entered in the journey log. It was recorded that the engine was operating at an N1speed of 2.3percent below the allowable limit of 93.3percent and an exhaust gas temperature (EGT) of 60C below the allowable limit of 590C. When the power-check calculations were reviewed during the investigation, it was determined that both were correct. Maintenance logs or personal notes did not record any results to verify the N1topping speed after the engine was installed into the helicopter before or after the N1adjustments. Such a record is not required by regulation. The engine power check recorded in the journey log provides a means of estimating the engine power output and tracking its performance. An N1topping check is required to confirm that the engine achieves the rated performance without exceeding its engine data plate placard limit. The pilot signed off the journey log, approving the aircraft for further flight. An engine vibration check was completed at the overhaul facility; however, a vibration check was not accomplished prior to returning the aircraft to service following the engine's installation in the helicopter. The Honeywell Lycoming engine maintenance manual specifies that a vibration check shall be performed following an engine installation. Conducting a vibration check provides the opportunity to verify that the engine-to-transmission alignment is correct and to detect any engine damage that could be the result of shipping or mishandling. The U.S. National Transportation Safety Board (NTSB) investigation report SEA85FA134 presents the following information retrieved from the manufacturer's T53 training manual: There is a common misunderstanding that it is safe to jackup the Takeoff Trimmer to permit N1 speeds approaching the overspeed limit of the compressor, and thereafter fly the aircraft by monitoring EGT and Torquemeter Pressure. The use of this procedure allows the engine to produce higher N1speeds and greater torque on a hot day, but the problem is that it also results in Turbine Inlet Temperatures (T5)higher than specification with consequential damage to components in the hot end. The Exhaust Gas Temperature (T9)measured on the T53Engine is not necessarily proportional to T5,and on a hot day it is possible to have T9within limits while T5is unknowingly too high. The manufacturer's Maximum Available Horsepower limits are designed to keep T5within specifications by reducing the available torque as the OAT [outside air temperature] increases. This is accomplished by keeping the Takeoff Trimmer adjusted to specification. If the trimmer is adjusted above specification, excessive T5temperatures will result in degradation of the Hot End components when the Engine is operating in hot weather. It is also noted that this procedure results in an increased risk of destructive compressor stalls (NTSBinvestigationFTW85FA188). The helicopter resumed work at the Bonaparte Lake site. After about 20minutes of working, the pilot relayed a message to the maintenance crew in Kamloops, requesting that the engine vibration test equipment be kept. Another 30minutes later, the pilot reported to the British Columbia Forest Service office that the helicopter was unserviceable and he returned to Kamloops. The pilot reported a buzzing cowling and that the engine lacked sufficient power to lift the load, resulting in the main-rotor rpm drooping.3 The vibration test kit (Chadwick-Helmuth7460A) was then installed on the engine and a flight test was carried out. The vibration check indicated that the engine was operating within specifications. The failure to perform a vibration check at initial installation, therefore, had no bearing on the subject failure, since a vibration check was conducted with no problems being indicated approximately 50flying minutes after engine installation. Subsequent to the test flight, the power and drooping issues were addressed by a further small increase in the N1take-off trim setting. No record of an N1topping check could be provided. The next morning (16August), the helicopter returned to the staging site and the pilot recorded 8.8hours of air time for the day, during which the pilot made several reports to indicate that operations were normal. No further adjustments were made to the engine or transmission when the helicopter returned to Kamloops in the evening. The helicopter was refuelled at Kamloops with 563litres of JetA (maximum capacity of 915litres). It is known that the N1speed was established to be 97.6percent at the overhaul facility, and this was the placard setting at which the engine produced its full-rated horsepower under International Standard Atmosphere conditions at sea level. By providing a physical restriction to fuel flow, the take-off trim setting is the N1overspeed protection device. It is possible to operate the engine at higher N1speeds, approaching the compressor overspeed limit, and it appears to produce all desirable effects; that is, more power, while apparently remaining within the allowable engine torque and EGT parameters (see textbox). Without the confirmation of an N1topping check after an adjustment, it is not known whether the N1speed is within specifications. After the N1was adjusted the last time, the pilot operated the helicopter the following day and made several reports indicating that operations were normal, which would suggest that the change appeared to be beneficial. However, if the N1was above specification, excessive T5temperatures could result in a degradation of the hot section components. 1.2 Pilot Information The pilot, experienced and familiar with the type of operation being conducted, had been hired by Gemini Helicopters Inc. on a temporary basis to provide relief for the pilot normally assigned to the helicopter. He possessed a valid Canadian airline transport pilot licence - helicopter, and had completed a pilot proficiency check on the Bell204 in June2003. Records indicate that the pilot had accumulated about 7500hours of total flight time, of which 200 hours had been flown on Bell204 helicopters in the previous 90days. He had flown C-GEAP during the last two weeks of June and then again beginning on 13August2003. 1.3 Meteorological Information The staging site was located at an elevation of about 4500feet above sea level (asl) in an area of rolling hills. The vegetation consisted of undergrowth in a primarily spruce forest. The prevailing weather conditions were sunny, hot and dry. At the time of the accident, a British Columbia Forest Service weather site, 14nautical miles to the southwest at an elevation of 3839feet asl, recorded a temperature of 20C with variable winds that reached a maximum speed of 8knots throughout the day. A normal lapse rate would yield a temperature at the elevation of the staging site of about 19C. An altimeter setting of 30.04inches of mercury was recorded at the Kamloops Airport, 33nautical miles to the south (elevation 1133feet asl). At an average working altitude of 4500feet asl, the weather conditions existing at the time equate to a density altitude of about 5900feet and a pressure altitude of about 4620feet. The previous day was about 5warmer, with an additional 3knots of wind. 1.4 Aircraft Information 1.4.1 General The Bell204B helicopter was equipped with a single Honeywell (Lycoming) T5311Bfree-turbine engine, with a design limit of 1100shaft horsepower. At the time of the accident, the helicopter had flown for about 10.9hours since the engine and transmission were installed. The helicopter was not equipped with a flight data recorder or a cockpit voice recorder, nor were they required by regulation. 1.4.2 Aircraft Performance The Bell 204B flight manual indicates that, under the ambient conditions that existed, the helicopter would hover out-of-ground effect at a gross weight of 3484kg. The estimated weight of the helicopter at the time of the accident was 3548kg. The collapsible Bambi water bucket had a maximum capacity of 1225litres (1225kg) and a maximum gross weight of 1280kg. To restrict the capacity, the bucket was equipped with a circumferential strap, which could be manually cinched. The bucket, as found, was cinched to 80percent capacity (980litres or 980kg). The helicopter gross take-off weight for external load operations was 4309kg. The centre of gravity was not considered to have been a factor. 1.4.3 External Load Operations The accident helicopter was equipped with a vertical reference kit that allowed for single-pilot operation from the left seat for longlining operations. Part of the vertical reference kit was a third foot pedal that provided mechanical release capability for the external cargo hook on the belly. A 100-foot braided fabric longline was in use and was equipped with an electrically operated (remote) hook assembly at the lower end, to which the water bucket was attached. Enclosed within the braid of the line was a pair of electrical wires to supply electrical power to operate either the remote hook or the bucket water-release mechanism. The wires were connected to the bucket water-release mechanism, thereby intentionally disarming the remote hook since it was not required for the operation or emergency purposes. Attached to the outside of the longline was a standard electrical extension cord, which would supply electrical power from a control box on the console between the pilot seats to operate a foam injection pump within the water bucket. The bucket was not equipped with such a pump and, therefore, this cord was not serving any purpose. Site examination of the external load system revealed that the water bucket was detached from the remote hook. There was no apparent damage to the remote hook, and subsequent testing determined that it could be operated manually and electrically. The control head of the bucket assembly with an attached shackle also remained intact and appeared to be undamaged. In general, helicopters have critical flight regimes from which the probability of establishing a successful autorotation is extremely low.4 The Bell204B Owner's Manual, Limitations section, specifies these conditions in its height and velocity diagram.5 The height and velocity diagram directs pilots to avoid operating under specified conditions of altitude and airspeed. However, this direction does not constitute a regulatory limitation for external load operations. Helicopters engaged in water-bucketing (external cargo) operations routinely operate under these critical conditions. Similar to other medium and large helicopters, there are five switches on the Bell204 cyclic control grip; however, the function of each switch can vary from one helicopter to another. Canadian Aviation Regulations (CARs)527.865 and 529.865require a primary quick-release system for the external cargo hook. However, neither the regulations nor standards specify that a particular switch must be assigned to the quick-release function for the external cargo hook. Therefore, there is no common design among manufacturers and operators to configure the function of these switches to suit their types of external load operations. Due to a previous pilot's preference, Gemini Helicopters Inc. had reconfigured three switches on the grip: the middle switch (activated by the thumb) operated the external cargo release, while either of the other two switches (the bottom switch activated by the little finger or the top conical switch, also operated by the thumb) operated the bucket water-release mechanism. Most of the pilot's experience consisted of longline operations, and records of the pilot's flight time indicate that at least 2700 hours of his most recent experience involved the use of cyclic grip configurations in which the bottom switch, operated by the little finger of the pilot's right hand, was used to release the external cargo hook. This helicopter was configured differently. Although the pilot knew that the external cargo hook release switch position had been changed, and he had used this system for two weeks in June and again for two days in August prior to the accident, it was not what he was accustomed to. Studies in human behaviour suggest that, amongst other variables, relative and finite amounts of practice influence which automatic behaviour occurs in an emergency situation; the more practised behaviour will be the default behaviour. The studies conclude that a pilot would require practice with a new switch configuration for 30days, or 85hours or 1000repetitions or more than with the known configuration, for it to become an automatic behaviour. With less practice, it would be difficult for the pilot to automatically and correctly select the appropriate switch to jettison the external load from the helicopter.6 The longline would normally be released from the external cargo hook in an emergency situation; however, the success of such an attempt may be hindered by the time available due to factors such as altitude, terrain, type of emergency and pilot familiarity with the specific release systems. In this occurrence, the longline was not released and it became entangled in a tree during the attempted emergency landing. It could not be determined what action the pilot took regarding the longline, nor was it possible to confirm the serviceability of either quick-release system. It was learned that the pilot's normal practice was to operate the helicopter with the external cargo hook switch in the armed position; however, damage prevented investigators from determining the position of this switch. The Transportation Safety Board of Canada (TSB) has identified reported occurrences where pilots have made errors in external-load release activations. Examples include the following: A02P0251 - Elaho Valley, Squamish, British Columbia, MD500D, 09 October 2002 A00W0020 - Robb Seismic Camp, Alberta, Bell 205A, 22 January 2000 A87P0025 - Gabriola Island, British Columbia, Bell 206, 15 May 1987 1.5 Wreckage and Impact Information 1.5.1 General The final heading of the helicopter was approximately 160 greater than the departure heading. The bucket became detached from the longline at the remote hook before impact; the longline with the remote hook remained attached to the helicopter as it approached the road. About 80feet from the bucket, in the direction of travel, the end of the longline with the remote hook was found wrapped around another tree that had broken off above, but the trunk remained standing. The longline lay in a direct line from this tree to the helicopter. The helicopter was found lying on its left side with all components of the airframe and thrust system present and in the correct orientation to each other except for the horizontal stabilizers, which were nearby but separated from the tail boom. The tail boom was adjacent to, but separated from, the main fuselage in the vicinity of the tail boom attachment point. The tail-rotor assembly, including the 90gearbox, was intact and in place on the vertical stabilizer, and the leading edges of both blades were devoid of impact damage. The tail-rotor driveshaft exhibited a torsion type of fracture most of the way around the circumference, a short distance ahead of the 42gearbox; both gearboxes contained oil and the pitch-link chain was intact. The main-rotor blades lay in a fore and aft orientation to the fuselage; the blades were bent and broken downward, but the leading edges were devoid of impact damage. The main-rotor control assembly lay in a normal orientation to the rest of the aircraft; however, the main-rotor mast was detached from the transmission. The transmission casing was completely melted, leaving the internal gears exposed. The transmission input quill was resting on molten transmission debris but remained attached to the engine by the main driveshaft. The input pinion gear was not broken. The aircraft cabin was severely damaged by the post-impact fire and little remained of the instrument or electrical panels. 1.5.2 Engine Examination Post-accident examination of the engine revealed that there was virtually no damage to the power turbine (PT) section, accessory gearbox or reduction gearbox of the engine; however, there was extensive damage to the gas producer section. The gas producer section consists primarily of five axial compressor disks (each with a spacer behind it) plus a single-stage centrifugal compressor, all driven by a single-stage compressor turbine (CT) at the exit of the combustion chamber. Surrounding each spacer, between the blades of sequential axial disks, is a set of stationary stator vanes attached to the outer casing of the compressor. During the engine teardown, it was observed that all blades on the fourth and fifth axial disks were fractured at or near the blade root. Twenty blade roots and 13disk posts of the fourth-stage disk were broken from the core of the disk itself. The debris, including the 20blade roots, was found piled up in the lower compressor case, half at the location of the fourth- and fifth-stage axial disks. Once the physical debris was removed from the compressor, the compressor rotor, including the compressor turbine, was able to rotate. The blade tips of the first three axial disks exhibited smearing of metal in both directions; the smearing damage was more pronounced with each stage from front to rear. The first, second, third and fourth stators and the exit guide vanes all had localized scoring, metal discoloration, and deformation on the outer portion of the forward and aft lip where contact was made with the spacers. Damage to the stators varied from localized cracking at the vane roots of the second stator to extensive gouging, tearing and forward deformation at the exit guide vane. All five spacers exhibited scoring damage where contact had been made between the spacers and the lip of the stationary stator vane assemblies. The aft end of the fourth-stage spacer (between the fourth and fifth axial disks) exhibited the greatest damage and had a score mark that was worn right through the spacer for about one-third of the circumference. The last stage of the compressor is the centrifugal impeller. Although there was significant scoring on the inlet of the impeller housing, damage to the impeller blades was primarily to the leading edges with only localized nicks and scoring further aft. The impeller housing (at the forward side of the impeller) exhibited depressions that corresponded to the shape of the impeller blades. Since there was no indication of smearing at the depressions, the TSB Engineering Laboratory concluded that the rotor assembly was likely not rotating at the time of ground impact. Run-out measurements of the third-, fourth- and fifth-stage spacers and the blade tips for the first three axial stages were taken before disassembly of the compressor rotor (the measuring device could not reach the first- and second-stage spacers). When plotted on a polar graph, the measurements of the blade tips displayed oval or elliptical shapes. The ovalized shapes became progressively more pronounced from the first stage to the third stage. Graphical plots of the measurements of the aft three spacers displayed round shapes centred away from the main axis of the rotor, with the maximum divergence at the fourth-stage spacer. The TSB Engineering Laboratory examined the disassembled compressor rotor components and confirmed that the blades on the first three axial stages varied in length around the circumference of each component and from component to component. The TSB Engineering Laboratory also concluded that there was no indication of a foreign object entering the compressor. All of the disk and blade fracture surfaces that were not smeared exhibited signs of overload failure; the balance of the fracture surfaces showed extensive smearing such that no conclusion could be reached with respect to the mode of failure. However, the condition of the fracture surfaces and the numerous pieces of blade material indicates that the rotor assembly continued to rotate for a period of time after the initial failure. At the discharge end of the combustion chamber, the hot gases pass through a stationary set of nozzle guide vanes (CT nozzle), which direct the gases onto the blades of the compressor turbine. After passing through the CT blades, the gases pass through another set of stationary nozzle guide vanes (PT nozzle) and then through the PT blades before exiting the engine. The distance between the CT and PT disks is approximately 5cm, with the PT nozzle occupying most of this space. It was observed that about one-half the CT nozzle vanes exhibited uneven trailing-edge damage indicative of melting. The ends and trailing edges of all CT blades exhibited the same type of damage and were reduced in size by approximately 30percent. The PT nozzle exhibited minor damage on the leading edges of the vanes, and there was virtually no damage to the PT blades. The compressor rotor assembly is supported at each end by a bearing; there were no indications of damage to either bearing. The CT disk (as found) was dynamically checked for balance and was found to be beyond specifications. Examinations for other indications of a CT wobble, which may have migrated forward through the number two bearing and applied an external bending force to the rigid axis of rotation of the compressor rotor, were carried out. Signs to support a progression of damage in this direction were not found. Such signs could be rub damage to the case surrounding the CT disk, damage to the number two bearing, contact between the hollow tail shaft of the compressor rotor and the power shaft rotating inside of it (in the opposite direction), or thermal damage to the PT nozzle and blades, suggesting a sustained over-temperature condition. The compressor rotor is a stacked-up assembly of several parts for which the manufacturer established a specific procedure detailed in the overhaul manual. Approved repair and overhaul facilities perform this work; Cappsco International Corp., which overhauled the engine, is approved by the U.S. Federal Aviation Administration (FAA) to complete this work and is subject to regular FAA audits. Regulatory requirements exist in the United States for maintenance facilities to administer quality control procedures and to provide qualified personnel.7 Records and procedures used for the overhaul work performed were reviewed. It was found that, during the compressor rotor assembly build-up procedure, the assembly was stacked vertically and began at the aft end. Once the second-stage disk was set in place, the power shaft and internal sleeve were installed, then the heated first-stage spacer was set in place, followed by the placement of the first-stage disk. The bolt holes in the first-stage disk must be aligned with bolt holes in the internal sleeve, and the manufacturer provides two guide pins to thread into the internal sleeve bolt holes to aid in aligning these holes.8 However, a hand-held punch was used for this task. This punch was then used in the attempt to rotate both pieces uniformly to align the internal tangs and slots of the first- and second-stage disks, which were obstructed from view by the sleeve. The result was that the tangs and slots of the first- and second-stage compressor disks were not aligned and did not mate correctly within the first-stage spacer. Indentations on the second-stage compressor disk (new part) indicated that at least two assembly attempts had been made to mate the first- and second-stage compressor disks while their respective slots and tangs were misaligned. In the process, one top corner of each tang was broken away. During the final compression, a piece of the removed material was crushed between the two disks to the dimension of the space remaining when both axial disks were pressed against the outer circumferences of the first-stage spacer. The TSB Engineering Laboratory found similar damage to the rear mating surface of the fifth-stage disk alignment slots and the impeller alignment tangs (also a new part). Prior to installing the sleeve, all tangs and slots could be inspected for proper engagement. Since the fit between the spacers and the disks is tight, the spacers are pre-heated to ease the assembly, placing some urgency on completing this task in a timely manner. The assembled compressor rotor assembly is then compressed in a hydraulic press. Three compression and release cycles are specified prior to applying the specified torque to the 10retainer bolts in the first-stage disk.9 Only one compression load was applied. The specific reason for three compression cycles or the consequences of using other than three cycles were not provided by the manufacturer. The length measurement of the compressor rotor assembly referred to in the overhaul manual as DimensionA is a comparison of three measurements taken around the circumference during the final compression cycle in the hydraulic press and again after removal from the press to determine that the plane of the first-stage disk is perpendicular to the axis of the assembly. DimensionA, taken after release from the press, is required to be recorded.10 This dimension was not taken while the rotor assembly was in the hydraulic press nor after it was released from the press. Another length dimension taken between the number one and number two bearing shoulders was taken and recorded, and was determined to be within specification. It was noted during the investigation that, when the dimensions of all applicable parts are summed up, the dimension between the number one and number two bearing shoulders falls within the specification without any compression of the assembly. The compressor rotor assembly was balanced prior to installation in the engine, and subsequent vibration checks, both in a test cell and during a test flight after installation in the helicopter, did not identify anomalies. Although the engine was not completely destroyed by the post-impact fire, it was exposed to enough heat to deform outer surfaces and to burn hoses and wiring bundles. It was decided that there was little chance of the interior gaskets and seals of the fuel control unit (FCU) remaining intact; therefore, the FCU was not tested or dismantled for examination. The position of the manual override (emergency) solenoid on the FCU, which can be selected by the pilot to bypass the power turbine governor (PTG) and control the fuel flow manually, was determined by X-ray to be in the automatic/normal position. The air supplied by the compressor provides for cooling as well as for combustion; therefore, reduced airflow resulting from any anomaly within the compressor itself may produce increased internal engine temperatures. As previously described, the compressor rotor blades exhibited indications of rub (contact between a rotor and its stator). Rub may typically be caused by mass imbalance, turbine or compressor blade fracture, defective bearings and/or seals, or by rotor misalignment, either thermal or mechanical.11 The transmission was also a new installation in the helicopter, and maintenance records were obtained regarding the overhaul and subsequent repair prior to installation. A detailed teardown examination of the transmission or any part of the drive train was not completed due to their condition and the fact that a drive train malfunction, in this model of helicopter, would not result in self-destruction of the power plant gas generator section. 1.6 Fire The source of ignition of the post-impact fire remains undetermined. The fuel tanks in the Bell204 are located at the lower-rear corners of the cabin structure; C-GEAP was fitted with large fuel tanks with a capacity of 915litres. The helicopter had been refuelled to capacity on the previous evening with Jet A grade turbine fuel. At the time of the accident, there would have been about 810litres of fuel on board.